D1 Agonist and D2 Antagonist Dual Effect of SPD 1433 Molecular dynamics simulations binding sites of the energy-minimum conformations of the DI and D2 receptors derived from the MD trajectories. The MD simulations were performed using the GrOMACS package ver sion3.1.4withtheGromoS96forcefield(26,27)(www.gromacs.org) The molecular topology file for SPD was generated with PRODRG (28) http://davapcl.bioch.dundeeac.uk/programs/prodrg/prodrg.html).Thepa RESULTS tial atomic charges of SPD were determined by using the CHelpG method (29) implemented in the Gaussian98 program(30)with the DFT/B3LYP/ 3D structures of D1 and D2 receptors unliganded D2, SPD-DI, and SPD-D2 complexes)were embedded in a Sequence alignment(Fig. S2 in the Supplementary Material) hydrated POPC lipid bilayer. The procedure and parameters for constructing indicates that the sequence identity and similarity are 21.8% the receptor/hydrated POPC systems are similar to those used in previous and 47.8%, respectively, for the TMs between rhodopsin membrane protein simulations(31-33). Fig. I shows, taking the SPD-D1 and the Dl receptor, 26.1% and 54.4% for the TMs between complex as an example, the structural mode of the receptor/PoPC/water rhodopsin and D2 receptor, and 44.5% and 66.4% between the four receptor/hydrated POPC systems first for all water molecules to the DI and D2 receptors themselves. Based on the high re their poor contacts with protein atoms, then for the whole system homology revealed by sequence alignments, the 3D models Im force was <10.00 kcal/mol- of the DI and D2 receptors were assembled taking the x-ray The solvent (water and PoPC) molecules of each initial system were crystal structure of bovine rhodopsin as a template.The PROCHECK (38)statistics showed that 90% of the residues molecules were constrained at 10. 50. 100. 200 and 298 K Afterward. each in both the DI and D2 models were in either the most favored system was equilibrated for 250 ps without any constraints To maintain the or in the additionally allowed regions of the ramachandran ystems at a constant temperature of 300 K, the Berendsen thermostat (34) map(Fig. S3 in the Supplementary Material), suggesting was applied using a coupling time of 0. 1 ps for the bulk water and PoPC. that the overall main chain and side chain structures are The values of the anisotropic isothermal compressibility were set to 45x all reasonable. The WHATIF (9)validation shows accept- 10-545x 10-5, 4.5 x 10-5,0,0, 0 bar"lfor xx, yy, zz, xy/yx, xz/zx, and able RMS Z-scores. All of these data suggest that the models yzizy components, respectively, for water and POPC simulations. The obtained through our homology modeling are reasonable lengths of all bonds, including those to hydrogen atoms, were constrained by Two kinds of interaction networks are observed in di and the LINCS algorithm(26). Electrostatic interactions between charged groups withing A were calculated explicitly, and long-range electrostatic interac- D2, 1.e, aromatic residue clusters(Fig. S4)and hydrogen tions were calculated using the particle mesh Ewald method (35)with grid bond (H-bond)network (Table SI). The presence of hydro- width of 1.2 A and a fourth-order spline interpolation. A cutoff distance of phobic cluster and H-bond interaction with TM Ill and TM 14 A was applied for the Lennard-Jones interactions. Numerical integratio vi in di could allow relative movement of tm vi. the of the equations of motion used a time step of 2 fs with atomic coordinates specific H-bonds cluster at the bottom of TM Ill and TM Vis saved every I ps for later analysis. To neutralize the modeled systems, 13, e added to the molecular systems of the free D1, D2, observed consisting of R-3.50 and E-6.30. It is conserved in both DI and D2 receptors, and the DRY motif of TM Ill may imulations were performed on these systems under the periodic boundary be relevant to the activation or inactivation of the receptor conditions in the nPt canonical ensemb The extracellular loop 1(EL-1)and extracellular loop 3 (EL-3)are typically short in all aminergic GPCRs. In con- trast,EL-2 is significantly longer and may reach into the active Molecular docking and binding energy calculation site crevice and form a lid over the bound ligand. There is a highly conserved disulfide bond between the conserved The geometry of the SPD ligand was built based on its crystal structure (6) Cys_e2 at the middle of EL-2 and Cy ys-3.25 at the beginning features the r configuration and a half-chair conformation of rings b and C. of TMIl. Thus in DI and D2 receptors, the stretches of only The dihedral angle between rings A and D is 1580. The molecular docking 4-5 residues between Cys_e2 and extracellular end of TM5 program AutoDock3. 05(37)was employed to probe the possible SPD. are in an extended state to reach two sides. It is noteworthy FIGURE 1 Overview of the SPD-DI/POPC/water sys- bilayer lipids are labeled H; and the lipid carbonyl oxygen atoms, choline N atoms, and P atoms are represented in a pace fill model (labeled P) and water as w: the 3-D size f the whole system was also labeled. The SPD in bindi Amplified binding SPD site was amplified and shown in a stick model. Biophysical Joumal 93(5)1431-1441Molecular dynamics simulations The MD simulations were performed using the GROMACS package ver￾sion 3.1.4 with the GROMOS96 force field. (26,27) (www.gromacs.org). The molecular topology file for SPD was generated with PRODRG (28) (http://davapc1.bioch.dundee.ac.uk/programs/prodrg/prodrg.html). The par￾tial atomic charges of SPD were determined by using the CHelpG method (29) implemented in the Gaussian98 program (30) with the DFT/B3LYP/ 6-311G** basis set. For MD simulations, the four models (unliganded D1, unliganded D2, SPD-D1, and SPD-D2 complexes) were embedded in a hydrated POPC lipid bilayer. The procedure and parameters for constructing the receptor/hydrated POPC systems are similar to those used in previous membrane protein simulations (31–33). Fig. 1 shows, taking the SPD-D1 complex as an example, the structural model of the receptor/POPC/water systems. Before MD simulations, energy minimizations were performed on the four receptor/hydrated POPC systems first for all water molecules to remove their poor contacts with protein atoms, then for the whole system until the maximum force was ,10.00 kcal/mol-A˚ . The solvent (water and POPC) molecules of each initial system were equilibrated with protein structures by constraining the solute (D1 or D2) at 300 K for 20 ps. Then the protein was equilibrated for 5 ps and the solvent molecules were constrained at 10, 50, 100, 200, and 298 K. Afterward, each system was equilibrated for 250 ps without any constraints. To maintain the systems at a constant temperature of 300 K, the Berendsen thermostat (34) was applied using a coupling time of 0.1 ps for the bulk water and POPC. The pressure was maintained by coupling to a reference pressure of 1 bar. The values of the anisotropic isothermal compressibility were set to 4.5 3 105 , 4.53 105 , 4.5 3 105 , 0, 0, 0 bar1 for xx, yy, zz, xy/yx, xz/zx, and yz/zy components, respectively, for water and POPC simulations. The lengths of all bonds, including those to hydrogen atoms, were constrained by the LINCS algorithm (26). Electrostatic interactions between charged groups within 9 A˚ were calculated explicitly, and long-range electrostatic interac￾tions were calculated using the particle mesh Ewald method (35) with a grid width of 1.2 A˚ and a fourth-order spline interpolation. A cutoff distance of 14 A˚ was applied for the Lennard-Jones interactions. Numerical integration of the equations of motion used a time step of 2 fs with atomic coordinates saved every 1 ps for later analysis. To neutralize the modeled systems, 13, 9, 12, and 8 Cl ions were added to the molecular systems of the free D1, D2, SPD-D1, and SPD-D2 complexes, respectively. Finally, four 10-ns MD simulations were performed on these systems under the periodic boundary conditions in the NPT canonical ensemble. Molecular docking and binding energy calculation The geometry of the SPD ligand was built based on its crystal structure (36) and optimized at the DFT/B3LYP/6-311G** level. This protonated structure features the R configuration and a half-chair conformation of rings B and C. The dihedral angle between rings A and D is 158. The molecular docking program AutoDock3.05 (37) was employed to probe the possible SPD￾binding sites of the energy-minimum conformations of the D1 and D2 receptors derived from the MD trajectories. RESULTS 3D structures of D1 and D2 receptors Sequence alignment (Fig. S2 in the Supplementary Material) indicates that the sequence identity and similarity are 21.8% and 47.8%, respectively, for the TMs between rhodopsin and the D1 receptor, 26.1% and 54.4% for the TMs between rhodopsin and D2 receptor, and 44.5% and 66.4% between the D1 and D2 receptors themselves. Based on the high homology revealed by sequence alignments, the 3D models of the D1 and D2 receptors were assembled taking the x-ray crystal structure of bovine rhodopsin as a template. The PROCHECK (38) statistics showed that 90% of the residues in both the D1 and D2 models were in either the most favored or in the additionally allowed regions of the Ramachandran map (Fig. S3 in the Supplementary Material), suggesting that the overall main chain and side chain structures are all reasonable. The WHATIF (39) validation shows accept￾able RMS Z-scores. All of these data suggest that the models obtained through our homology modeling are reasonable. Two kinds of interaction networks are observed in D1 and D2, i.e., aromatic residue clusters (Fig. S4) and hydrogen bond (H-bond) network (Table S1). The presence of hydro￾phobic cluster and H-bond interaction with TM III and TM VII in D1 could allow relative movement of TM VI. The specific H-bonds cluster at the bottom of TM III and TM V is observed consisting of R-3.50 and E-6.30. It is conserved in both D1 and D2 receptors, and the DRY motif of TM III may be relevant to the activation or inactivation of the receptor. The extracellular loop 1 (EL-1) and extracellular loop 3 (EL-3) are typically short in all aminergic GPCRs. In con￾trast, EL-2 is significantly longer and may reach into the active site crevice and form a lid over the bound ligand. There is a highly conserved disulfide bond between the conserved Cys_e2 at the middle of EL-2 and Cys-3.25 at the beginning of TM III. Thus in D1 and D2 receptors, the stretches of only 4–5 residues between Cys_e2 and extracellular end of TM5 are in an extended state to reach two sides. It is noteworthy FIGURE 1 Overview of the SPD-D1/POPC/water sys￾tem. D1 is shown in ribbon; the hydrophobic chains of bilayer lipids are labeled H; and the lipid carbonyl oxygen atoms, choline N atoms, and P atoms are represented in a space fill model (labeled P), and water as W; the 3-D size of the whole system was also labeled. The SPD in binding site was amplified and shown in a stick model. D1 Agonist and D2 Antagonist Dual Effect of SPD 1433 Biophysical Journal 93(5) 1431–1441